A Comparison of Wind Turbine Aeroelastic Codes Used for Certification

Transcription

1 National Renewable Energy Laboratory Innovation for Our Energy Future A national laboratory of the U.S. Department of Energy Office of Energy Efficiency & Renewable Energy A Comparison of Wind Turbine Aeroelastic Codes Used for Certification Preprint Conference Paper NREL/CP January 26 M.L. Buhl, Jr. National Renewable Energy Laboratory A. Manjock Germanischer Lloyd WindEnergie GmbH To be presented at the 44 th AIAA Aerospace Sciences Meeting and Exhibit Reno, Nevada January 9 12, 26 NREL is operated by Midwest Research Institute Battelle Contract No. DE-AC36-99-GO1337

2 NOTICE The submitted manuscript has been offered by an employee of the Midwest Research Institute (MRI), a contractor of the US Government under Contract No. DE-AC36-99GO1337. Accordingly, the US Government and MRI retain a nonexclusive royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for US Government purposes. This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available electronically at Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy Office of Scientific and Technical Information P.O. Box 62 Oak Ridge, TN phone: fax: mailto:[email protected] Available for sale to the public, in paper, from: U.S. Department of Commerce National Technical Information Service 5285 Port Royal Road Springfield, VA phone: fax: [email protected] online ordering: Printed on paper containing at least 5% wastepaper, including 2% postconsumer waste

3 A Comparison of Wind Turbine Aeroelastic Codes Used for Certification Marshall L. Buhl Jr. * National Renewable Energy Laboratory, Golden, Colorado 841 and Andreas Manjock Germanischer Lloyd WindEnergie GmbH, Hamburg, D-2459, Germany The National Renewable Energy Laboratory (NREL) has created aeroelastic simulators for horizontal-axis wind turbines. The U.S. wind energy industry often uses these codes to analyze its turbines, and manufacturers seeking type certification for their turbines need to use codes that certifying agencies will accept. NREL brokered an agreement with Germanischer Lloyd (GL) WindEnergie GmbH, the world s foremost certifying body for wind turbines, to participate in a comparison between its code and two of NREL s codes. After a variety of exercises, GL issued a statement that it is acceptable for manufacturers to use the NREL codes for on-shore wind turbine certification. I. Introduction HE National Renewable Energy Laboratory (NREL) and its academic and industry partners have created T aeroelastic simulators for horizontal-axis wind turbines (HAWTs). Many members of the U.S. wind energy industry use these codes to estimate the design loads of their turbines. When manufacturers seek type certification for their turbines, they need to use codes that the certifying agencies will accept. The world s foremost certifying body for wind turbines is Germanischer Lloyd (GL) WindEnergie GmbH, located in Hamburg, Germany. NREL and GL agreed to participate in a comparison of their codes. The goal of this agreement was for GL to test the accuracy of the NREL codes and, if found acceptable, issue a statement to that effect. After the comparison, GL issued this statement and a report on the study. 1 2 (Fatigue, Aerodynamics, Structures, and Turbulence) is NREL s primary aeroelastic wind turbine simulator. It is a medium-complexity, turbine-specific code that models HAWTs with two or three blades. It uses NREL s AeroDyn 3,4 subroutine library of rotor-aerodynamics routines to compute the aerodynamic forces on the turbine blades. MSC.Software developed 5, 6 (Automatic Dynamic Analysis of Mechanical Systems) as a commercial, multibody-dynamics code. Users can model a wide variety of structures with this code. NREL linked the AeroDyn 78 library with to enable it to model HAWTs. By means of pre-processing logic, generated the input files for the models. These models included features not available in, even though created them. One important additional feature is the blade torsion degree of freedom. NREL has made extensive comparisons 9 11 between and. Because they share aerodynamics routines, these studies mostly compared the structural response. In the most recent comparison, the two codes agree extremely well for a wide variety of turbines, from small to large. Differences appear when the models use features not available in, such as blade torsional flexibility. GL s code is (Dynamic Analysis of Horizontal Axis Turbines). It, too, is a HAWT-specific code with a similar design philosophy to. To evaluate the NREL codes, NREL and GL agreed to model two turbines. Different types of turbines were required to exercise different features of the codes. To publish the results of the study without constraint, the participants chose two public domain turbine designs. The first turbine was the 1-m, two-bladed, stall-regulated turbine * Senior Engineer II - Mechanical, National Wind Technology Center, NWTC/3811, 1617 Cole Blvd., Golden, CO, USA, ASME Member. Expert In Charge, Load Assumptions Department, Steinhöft 9, D-2459, Hamburg, Germany. 1

4 used in the NREL s Unsteady Aerodynamics Experiment (UAE) conducted in the NASA Ames 8'x12' wind tunnel. Although the UAE rotor is capable of teetering, this study used only the upwind configuration with locked teeter. The second turbine was a fictional concept turbine designed by Global Energy Concepts and Windward Engineering for NREL s WindPACT (Wind Partnership for Advanced Component Technologies) program. The WP 1.5 is a three-bladed, upwind turbine with a 7-m rotor with active pitch control and a variable-speed generator. We started with existing models of the two turbines. s preprocessor generated our models from the input files. GL used the properties in the input files to create its model. Although the project started with simple comparisons, it became more complex as the participants gained confidence that the codes used the same properties for their models. They compared aggregate properties such as mass, centers of mass, inertias, and natural frequencies. The first simulations were for rigid turbine models rotating in a vacuum. Other simulations were for fully flexible turbines that included an abrupt wind drop for parked turbines, yawing at a fixed rate with the rotor locked, unsheared gusts, and 1-minute simulations with full-field, stochastic turbulence. The 1.5-MW WindPACT turbine, with its relatively low rigidity in blade torsion, showed the importance of being able to model that degree of freedom (DOF) and the blade center-of-mass offsets that help drive it. Another finding is the importance of applying the aerodynamic forces at the deflected orientations of the blades instead of the undeflected orientations. The primary purpose of this study was to determine whether the NREL codes are accurate enough for use in turbine certification. Because of time and funding constraints, the project included only code-to-code comparisons, not validating comparisons to test data. This paper will provide a detailed description of the aeroelastic simulators, the turbines modeled, and the procedures used in the project. It will include comparisons of key loads for the various simulations and illuminate the strengths and weaknesses of the various codes and the importance of modeling advanced blade properties. II. The Codes A. is a sophisticated, commercial, multibody-dynamics simulator that can model virtually any turbine type if coupled with an appropriate fluid-dynamics model. MSC.Software develops and licenses it. We used 23 for this study. is not a wind-turbine-specific code, and many use it to analyze other types of systems such as cars, robots, and spacecraft. It is a well tested program, and we believe the rigid-body predictions of this fully nonlinear code are very accurate. uses lumped masses connected by flexible fields similar to multidimensional spring dampers to model flexible structures. The flexible fields use some approximations and are not as exact as the rigid-body portion of the simulator. As distributed by MSC.Software, has no built-in rotor-aerodynamics algorithms. However, it allows users to link their own routines with it. NREL links its AeroDyn aerodynamics module with to model the propeller-type rotor aerodynamics. is a complex program, and it normally requires a large effort to model a wind turbine with it. To simplify and speed this process, NREL added -preprocessing capabilities to. We used the -to- preprocessor to construct all models used in this study. Because of the way models flexible parts of the structure, it has capabilities unlike the other codes used in the study. Turbine-specific codes typically have just a few DOFs to model the blades and tower. models usually use more than 1 DOFs for each flexible substructure. This includes not only bending modes but also torsional and extensional modes. can model the low-speed shaft as a fully flexible structure to model rotor whirling motion, but we limited it to only a torsional DOF to mimic the other codes. also allows users to specify the location of the elastic axis and the centers of mass of cross-sections. Results from this study show that these offsets and live twist are important in modeling flexible turbines. B. GL s code is a "traditional" wind-turbine-specific code. Its development began in the late 198s as a quasi-stationary, rigid, structural analysis tool. With changing demands, the code evolved to include aeroelastic coupling and a better description of the wind turbine. The code is still under development to meet evolving needs. 2

5 The aerodynamics calculations in use the classic, blade-element/momentum (BEM) theory from Wilson or Politis 12,13 to compute induced velocities. Now dynamic and turbulent wake conditions, 14 as well as dynamic stall, are available options. The dynamic stall models used are from Stig Oye 15,16 or Goman and Khrabrov 17. The code models blade, drive train torsion, and tower flexibility. Blade-bending flexibility through modal representation is independently analyzed for every blade and considers full edge-flap coupling. Up to six modes for every blade can be considered, but usually only the first two or three modes are used. models the drive train as a simple two-mass spring-damper system with input torque from the hub and the generator. The generator can be asynchronous or fully controlled synchronous. The yaw bearing is a brake, which can slip, and a spring damper system. The tower model used in this study considered only fore-aft and side-to-side bending, but a new version of in testing includes tower torsion, as well as the influence of linear springs at the different nodes to model guy wires. A set of different parametric pitch-controller algorithms is available representing typical wind turbine controller designs that can be adjusted by a few parameters such as proportional-integral-derivative (PID) omega/pitch control and proportional-integral power/pitch control. comes with a package of pre- and post-processing tools. The turbulent wind for the input is generated by a turbulence generator using an ESDU (modified von Karman) spectrum delivering a three-dimensional, three-component, and anisotropic wind field. Another tool computes the fatigue-analysis of calculated time series, a program using the rainflow counting method, including weighting of the results according to given wind distributions. Further, a huge number of statistical analysis functions are included within the tool, which cover the requirements of detailed load analysis and load assessment procedures. We used v19 (Sept. 21) for this study. C. Like, is a wind-turbine-specific code. Therefore, it has limited DOFs but can model many common turbine configurations. also models flexible elements using modal representation. The reliability of this representation depends on the generation of accurate mode shapes, which are input into. We used a program called Modes 18 to generate these shapes. Like, Modes originally came from Oregon State University, but NREL rewrote most of it. The blade and tower models also use properties such as stiffness and mass per unit length to specify the flexibility characteristics. As with, uses the AeroDyn module of routines to model the rotor aerodynamics. We used v5.1 in this study and added a custom-written PID pitch controller for the WP 1.5 turbine. This version of allows up to 18 DOFs. These DOFs include two blade-flap modes and one uncoupled blade-edge mode per blade. It also has two fore/aft and two side-to-side tower bending modes. It has drive train torsional flexibility similar to and a generator DOF with multiple models. We used the simplest model for the UAE turbine in this study. For the WP 1.5 turbine, the control system computed the desired generator torque, so we did not use any of the available generator models. has nacelle yaw, rotor teeter for two-bladed turbines, and hinges between the rotor and tower and between the tower and tail for furling turbines. Of this last group, we used only nacelle yaw in this study and for only one test case. For the yaw model, can allow it to rotate freely or can apply a spring/damper. For most of this study, all the codes disabled the yaw DOF. Comparisons between and an earlier version of (called _AD) illuminated many weaknesses in, so it received a major overhaul that began in 22. The latest version of shows excellent agreement with. 11 An attractive feature of is that it is a public domain code freely available to anyone. Its distribution archive includes the full source code for the program. The source code allows users to know what the program is doing, and it allows them to modify the code to suit their needs. D. AeroDyn AeroDyn is part of a simple wind turbine aeroelastic simulator called YawDyn 19. The University of Utah developed YawDyn under a subcontract with NREL. When NREL engineers started using, they needed routines to compute aerodynamic forces to apply to the structure. NREL contracted with the Utah group to extract the aerodynamic routines from YawDyn so that could use them. That effort resulted in the AeroDyn subroutine library. The principals from the University of Utah formed an independent company called Windward Engineering, and that organization now supports AeroDyn and YawDyn for NREL. 3

6 AeroDyn is a library of subroutines that can be linked with structural codes to provide the lift, drag, and pitching-moment forces on the blade. To calculate the induced velocities, AeroDyn has two options: BEM theory based on the PROPX code 2 and generalized dynamic wake (GDW), which uses Peters and He s method. 21 The BEM induction can optionally include tip and hub losses based on Prandtl s work as described by Glauert. 22 AeroDyn also has an option to use dynamic stall based on the work of Leishman and Beddoes AeroDyn can excite the rotor structure with steady or turbulent, sheared, hub-height winds, or by using full-field stochastic winds. We used NREL s TurbSim 26 v1.c-bjj to generate the full-field turbulent winds for the AeroDyn-based codes in this study. The version of AeroDyn used in this study has a weakness in the GDW algorithm that causes numerical instabilities at low wind speeds when the induction is large (turbulent windmill state and vortex-ring state). Because of this, AeroDyn would automatically switch to BEM analysis whenever the mean wind speed of simulations using full-field turbulence was 8 m/s or less. The latest version of AeroDyn is numerically stable throughout the turbine operating range. For this study, we linked AeroDyn v12.57 with and with to create their executable files. III. The Turbines Modeled A. Unsteady Aerodynamics Turbine The first turbine to be modeled was used by NREL in the UAE tests performed in the NASA Ames 8'x12' wind tunnel in May 2. Test engineers had the option to allow the two-bladed rotor for the UAE turbine to teeter or be locked. Almost all commercial turbines have three blades, and GL s has no teeter DOF, so we kept the UAE rotor rigid for this evaluation. The turbine also had the capability to run upwind with locked yaw or downwind with free yaw. Virtually all commercial turbines run upwind, so we modeled it this way in this study. NREL designed the UAE turbine to be extremely stiff so that it could be used in the aerodynamics studies. Because it is so stiff, the blades twisted very little in the model, so there was excellent agreement between the and predictions for the UAE turbine. Properties used for the UAE model came from the report documenting the wind tunnel test 27. We list some of the more important properties in Table 1. The UAE model for cases with full flexibility used 21 analysis nodes for each of the two blades and for the tower. The blades used 21 flexible segments to join the nodes, which resulted in 126 (6x21) DOFs for each blade. The tower used 22 flexible segments to join the 21 nodes and the tower top, which resulted in 132 (6x22) DOFs. The model also included drive-train-torsion and generator DOFs. The full system had 386 DOFs. Table 1. UAE Turbine Properties. Number of Blades 2 Rotor Diameter 1 m Hub Height 12.2 m Rated Power 2 kw Fixed Speed 72 rpm Stall Control Upwind Fixed Yaw Rigid Hub B. WindPACT 1.5-MW Baseline Turbine The other turbine we modeled was the baseline paper turbine used in NREL s WindPACT Turbine Rotor Design Study 28. That fictional turbine is a typical, utility-scale turbine rated at 1.5 MW. We had originally intended to use a commercial turbine for the second turbine, but the logistics of passing proprietary data among all parties became onerous, and we abandoned that plan. NREL attempted to replace the state-space pitch-control system previously used by NREL for this turbine with a simpler PID pitch-control system for the and models. However, after the completion of the study, NREL discovered that the control system was not a true PID pitch-control system. It was more of a pitch-rate control system, and it commanded the pitch change that would be accomplished in one time step at the desired pitch rate. The model had no pitch-actuator model, and it set the actual pitch angle to the commanded pitch angle. When generated the WP 1.5 model, it created an actuator model unlike the one used in the original WindPACT study. NREL found that tuning the spring and damping constants for the new WP 1.5 actuator model 4

7 was problematic. GL used their built-in PID control system and tuned it to approximate the NREL control system. After a defined time lag, sets the pitch angle to the commanded pitch setting. For this turbine, we attribute many of the differences between the simulators to the differences in the control systems. Because of some of the changes NREL made to the WP 1.5 control system, the models predicted a maximum power of 1.2 MW instead of the original 1.5 MW. In the interest of time, we decided to accept the lower maximum power as we were not comparing results to a real turbine. Properties for the WP 1.5 model came from the model used in the WindPACT study. We list some of the more important properties in Table 2. The WP 1.5 model for cases with full flexibility used 15 analysis nodes for each of the 3 blades and 1 nodes for the tower. The blades used 15 flexible segments to join the nodes, which resulted in 9 (6x15) DOFs for each blade. The tower used 11 flexible segments to join the 1 nodes and the tower top, which resulted in 66 (6x11) DOFs. The model also included drive train torsion and generator DOFs. The full system had 338 DOFs. Table 2. WP 1.5 Turbine Properties. Number of Blades 3 Rotor Diameter 7 m Hub Height 84.3 m Rated Power 1.5 MW Variable Speed Cut-in to rated wind speed Variable Pitch Rated to cut-out wind speed Upwind Yaw Control Rigid Hub IV. Test Cases A. Vacuum Cases We used the first two tests to determine if we had correctly modeled the basic configuration and mass properties. For the first test, we disabled all but the yaw DOF and yawed the turbine at a steady rate in a vacuum. The rotor did not rotate during this test. In general, the agreement was quite good, and the differences were in the noise level. For the second case, we enabled all DOFs but yaw while holding the rotor speed constant in a vacuum. See Fig. 1 for a UAE turbine example. 2 Blade-Root IP Bending Moment, kn m Figure 1. UAE blade-root in-plane bending moment for the case of a rigid, rotating rotor in a vacuum. Because of a lack of aerodynamic damping, there was some oscillation of the flexible parts for the WP 1.5 turbine, which masked the comparisons. However, the load levels were in good agreement. 5

8 B. Abrupt Wind Drop For this case, we modeled the turbines with full flexibility. We parked the rotor and locked the yaw. We set the blades so they were flat to the wind. This was the first case that included aerodynamics effects. We blew a steady, uniform 6 m/s wind on the turbines and then dropped the wind speed to zero instantaneously. The WindPACT turbine has flexible blades. In this case, the wind bent the blades back quite dramatically by several meters. As seen in Fig. 2, all codes agreed. and agreed perfectly despite their fundamentally different methods of modeling flexibility. uses the lumped mass method, while the other two codes use defined modes. and share identical aerodynamic routines, which eliminates a significant potential difference. The small differences to during wind load period are probably the result of differences in the aerodynamic routines. 5 Blade-Root OoP Bending Moment, kn m Figure 2. WP 1.5 blade-root out-of-plane bending moment for the Abrupt Wind Drop case. This case illuminated a difference in the way aerodynamic loads are applied to the blades. All codes apply the aerodynamic forces at the deflected position of the blade. However, only and apply these forces in the deflected orientation (Fig. 3). In older-technology turbines with stiffer blades, this was not a modeling problem. However, as turbines become larger and more flexible, the deflected orientations become more important. This difference between modeling methods shows quite dramatically in the radial loads of the blade root, as one can see in Fig. 4. applies the aerodynamic forces perpendicular to the undeflected blades. Thus, the only axial force acting on the blade is gravity. This is probably the cause of the small differences in the out-of-plane bending moments seen in Fig. 2. 6

9 Radial Component, Undeflected Blade Deflected Blade Axis of Rotation Figure 3. Application of aerodynamic forces to a bent blade. Blade-Root Radial Force, kn Figure 4. WP 1.5 blade-root radial force for the Abrupt Wind Drop case. 7 When we look at the yaw-bearing rolling moment (Fig. 5), we see that the model gives much higher values when the wind is blowing. Because the rotor is not turning and the simulators used only simple aerodynamics, the likely cause of the difference is blade torsion. The tip rotation in the model was over six degrees. Neither of the other codes has this DOF. 7

10 Yaw-Bearing Rolling Moment, kn m Figure 5. WP 1.5 yaw-bearing rolling moment for the Abrupt Wind Drop case. Figure 6 shows the blade-root pitching moment. doesn t compute blade pitching moments nor account for the blade torsion due to aerodynamic and inertial forces acting on a bent blade, so it predicts a zero value. and are close, and their difference is likely due to blade twist. Blade-Root Pitching Moment, kn m Figure 6. WP 1.5 blade-root pitching moment for the Abrupt Wind Drop case. 7 C. Uniform Gust This case modeled the turbines with full flexibility. We superimposed a 1 m/s sinusoidal gust on a steady 15 m/s wind. There was no change in wind direction and no wind shear. This case was the first to highlight a significant shortcoming in the way we modeled the WP 1.5 turbine with. The model included no pitch actuator model, and the actual pitch was set to the commanded pitch. The model did include a pitch actuator, and the default values chosen prevented the blade pitch from slewing at even one deg/s (Fig. 7). We did not discover this problem until after we ran all the test cases and began writing the report. As can be seen in the plots of rotor power (Fig. 8), the slow pitch actuator had a significant effect on all system loads. 8

11 Blade Pitch, deg Figure 7. WP 1.5 blade pitch for the Uniform Gust case. 25 Rotor Power, kw Figure 8. WP 1.5 rotor power for the Uniform Gust case. D. Full Models Operating in Turbulent Wind 1. Description The series of simulations used to model the turbines in normal operation with stochastic winds was the decisive test for the comparisons. We used mean wind speeds from 5 25 m/s in steps of 2 m/s. We used the IEC von Karman model with the A turbulence intensity as specified in the IEC standard. 29 The models we used for this series included all the appropriate DOFs, and the model used the 1.2-MW rated power for the WP 1.5 that the NREL codes used. For this case, we not only compared time series, but we also calculated damage-equivalent loads (DELs) and cumulative fatigue spectra. To define the annual operation conditions, we used a Rayleigh wind-speed distribution with an 8.5 m/s mean wind speed and assumed a 2-year lifetime for the turbines. For the DEL calculations, we set the number of constant-load cycles to 1 million. We used an S/N slope of 1 for composite parts and 4 for steel parts. 2. UAE Results For the UAE turbine, we documented only the resulting power curve from the series of simulations. We computed the mean rotor power for each case and plotted the results, which show that and agree exactly (Fig. 9). The fact that they share common aerodynamics routines from AeroDyn make such excellent agreement relatively easy to achieve. The structural theory base for the two codes is completely different, but it appears we have successfully represented the turbine in both simulators and that elastic deflections are not significant. The results of the model showed a small deviation from the other two codes. We suspect that the disagreement is 9

12 inherent in the aerodynamics algorithms used by the codes. At low wind speeds, the NREL codes used BEM instead of GDW because of numerical instability problems in the GDW algorithm at low wind speeds. uses a different dynamic-stall model than AeroDyn, which probably leads to the deviation in post stall power production in turbulent wind. 15 Rotor Power, kw Figure 9. UAE rotor power curve. 3. WP 1.5 Results At 9 m/s and below, and are in perfect agreement. However, above that point, the problems with the pitch actuator cause to diverge significantly from the and results. Except for the difference at 9 m/s, and agree quite well (Fig. 1). 15 Rotor Power, kw Figure 1. WP 1.5 rotor power curve. We computed cumulative fatigue spectra for all loads channels. As seen in the comparison of flap-bending moments in Fig. 11, the three codes grouped quite nicely. The chart of root radial forces shows that and agree well, while predicts quite different results (Fig. 12). The difference is due to the low pitch rate available in the model, which becomes a problem at higher wind speeds. The model required about 5 s to reach the appropriate pitch setting of 3 deg (Fig. 13), so the loads were very high during the first part of the run. Once the pitch reached equilibrium, agreed well with the other codes, which can be seen in the spectra above load cycles. 1

13 3 Blade-Root Flap Bending Moment, kn m 2 1 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8 1E+9 1E+1 Cumulative Fatigue Cycles Figure 11. WP 1.5 blade-root flap bending moment cumulative spectra. 3 Blade-Root Radial Force, kn 2 1 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8 1E+9 Cumulative Fatigue Cycles Figure 12. WP 1.5 blade-root radial force cumulative spectra. 4 Blade Pitch, deg Figure 13. WP 1.5 blade pitch for 25 m/s mean wind speed. 11

14 There is a small difference in the curves for the shaft-bending moments in Fig. 14. Various channels tell similar tales. In general, the codes agree with each other. However, the downwind thrust on the yaw bearing (Fig. 15) reveals considerable discrepancies. When looking at the 25 m/s time series for the NREL codes, it is apparent that both codes had not reached equilibrium when the recording started at 1 s. This caused a huge jump in the cumulative spectra in the low-cycle region. A probable cause of the long startup transient is that AeroDyn uses BEM for the first second of a GDW simulation. NREL eliminated this use of BEM for 1 s in the latest version of AeroDyn. This switchover causes the blade pitch to be at the wrong setting when GDW takes over. It seems that the transient took longer than 9 s to die out. With the tower shaking back and forth violently during the transient, the combined rotor and nacelle mass (thrice that of the rotor) amplified the forces at the tower top more than it did at the rotor, so the difference is not so pronounced in the blade and rotor loads. s equation-of-motion solver generally finds an equilibrium within the first simulation steps, thus the transient period was much shorter compared to and. This explains the typical spectral differences seen for the rarer, low-cycle events. 3 Shaft Bending Moment, kn m 2 1 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8 1E+9 1E+1 Cumulative Fatigue Cycles Figure 14. WP 1.5 shaft-bending moment cumulative spectra. Yaw-Bearing Downwind Thrust, kn E+3 1E+4 1E+5 1E+6 1E+7 1E+8 1E+9 1E+1 Cumulative Fatigue Cycles Figure 15. WP 1.5 yaw-bearing downwind force cumulative spectra. Conclusion This paper contains a small selection of results from the codes,, and. After a profound study of the comparisons between the codes based on all components relevant for wind turbine design and certification, GL decided that the differences in the predictions are within reason, so they issued a certificate of acceptability. 12

15 We believe that modeling errors, such as the pitch actuator used for, caused many of the errors, but that differences in the feature sets of the various codes also had a significant impact. Only has a blade torsional DOF, and it has full coupling between the flap, edge, and torsional modes. Another unique feature is blade section center-of-mass offsets. and apply aerodynamic forces at the deflected orientations of the blades, while applies them in the original orientations. During normal operation when blade deflection is moderate, it is a reasonable compromise. However, for computing the extreme loads of a parked turbine, the approximation may not be valid for specific components. All of these codes have strengths and weaknesses. A great benefit from studies such as this one is that they help us decide which features we want to work on next in the continual development of our codes. After a decade of code comparisons performed by NREL, much progress has been made on both sides of the Atlantic. The comparisons have left some items still to be clarified and reveal the need for some additional effort. We would like to fix the modeling errors to help us isolate the real differences among the codes. We would try things such as setting the blade torsional stiffness much higher in to determine whether that was the source of some of the differences. We would investigate setting the blade-section centers of mass to align with the pitch axis for the models and observe that effect. Similarly, setting the aerodynamic centers to align with the pitch axis might tell us if those offsets caused some differences between NREL s codes and. NREL hopes to work with GL to retest,, and AeroDyn every other year as they develop new versions of the codes. They plan to replace the UAE turbine with the NREL Offshore Baseline 5-MW turbine for the next comparison. Like the WP 1.5, it is a fictional turbine used by NREL for code-development efforts. By including an offshore turbine in the comparison, NREL hopes to convince GL that its codes are also acceptable for analysis of offshore turbines. Acknowledgments M. Buhl thanks Christian Nath and Axel Andreä at GL for making it possible for NREL to gain acceptance of its codes. He also thanks Kimon Argyriadis of GL for describing for this paper and answering questions about it while A. Manjock was away. Sandy Butterfield of NREL was the man with the vision of achieving GL s acceptance of the NREL codes. Thanks also to Jason Jonkman of NREL for turning into the excellent design code it is today. Jason and Gunjit Bir (also of NREL) helped M. Buhl interpret some of the results. Katie Johnson (formerly of NREL) and Alan Wright of NREL helped M. Buhl design the control system for the WP 1.5 turbine. We would also like to thank Ruth Baranowski at NREL for making this a much easier report to read. The authors performed this work at GL and NREL in support of the U.S. DOE under contract number DE- AC36-99-GO1337. References 1 Manjock, A., Evaluation Report: Design Codes and for Load Calculations of Onshore Wind Turbines, Germanischer Lloyd WindEnergie GmbH, Rept. 7242, Hamburg, Germany, May Jonkman, J. M. and Buhl Jr., M. L., " User's Guide," NREL/EL , Golden, CO: National Renewable Energy Laboratory, Laino, D. J. and Hansen, A. C., User's Guide to the Computer Software Routines AeroDyn Interface for, Salt Lake City, UT: Windward Engineering, Prepared for the National Renewable Energy Laboratory under Subcontract No. TCX , September Moriarty, P. J. and Hansen, A. C., AeroDyn Theory Manual, NREL/TP , Golden, CO: National Renewable Energy Laboratory, August Elliott, A. S. and McConville, J. B., Application of a General-Purpose Mechanical Systems Analysis Code to Rotorcraft Dynamics Problems, Prepared for the American Helicopter Society National Specialists Meeting on Rotorcraft Dynamics, Elliott, A. S., Analyzing Rotor Dynamics with a General-Purpose Code, Mechanical Engineering, Vol. 112, No. 12 (December 199): pp Laino, D. J. and Hansen, A. C., User's Guide to the Wind Turbine Dynamics Aerodynamics Computer Software AeroDyn, Salt Lake City, UT: Windward Engineering, Prepared for the National Renewable Energy Laboratory under Subcontract No. TCX , December Moriarty, P. J. and Hansen, A. C., AeroDyn Theory Manual, NREL/TP , Golden, CO: National Renewable Energy Laboratory, August Buhl, M. L., Jr., Wright, A. D., and Pierce, K. G., "Wind Turbine Design Codes: A Comparison of the Structural Response," Collection of the 2 ASME Wind Energy Symposium Technical Papers at the 38th AIAA Aerospace Sciences Meeting and Exhibit, 1 13 January 2, Reno, NV, AIAA and ASME, New York, February 2, pp

16 1 Buhl, M. L., Jr., Wright, A. D. and Pierce, K. G., "_AD Code Verification: A Comparison to," Collection of the 21 ASME Wind Energy Symposium Technical Papers at the 39th AIAA Aerospace Sciences Meeting and Exhibit, January 21, Reno, NV, AIAA and ASME, New York, February 21, pp NWTC Design Codes ( Verification by Jonkman, J.; Buhl, M. L., Jr.). URL: simulators/fast/verification. Last modified May 19, 25; accessed September 21, Wilson, R. E., Applied Aerodynamics of Wind Power Machines, Corvalis, Oregon: Oregon State University, July Politis, G. K. and Loukakis, T. A., On Lifting Line Analysis of Horizontal Axis Windturbines, Wind Engineering, Vol. 8, No. 1, 1984, pp Meyer, M. and Matthies, H. G., Strukturdynamik und Dynamic Stall Zustandsraum-Darstellung der instationären aerodynamischen Beiwerte, Erneuerbare Energien, Vol. 3, Øye, S., Dynamic Stall Simulated as Time Lag of Separation, Proceedings of 4 th IEA Symposium on Aerodynamics of Wind Turbines, IEA, Rome, ETSU-N-118, November Øye, S., Revised Dynamic Stall Model, AFM Report 95-6, Lyngby, Denmark: Technical University of Denmark, November Gorman, M. and Khrabrov, A., State-Space Representation of Aerodynamic Characteristics of an Aircraft at High Angles of Attack, Journal of Aircraft, Vol. 31, No. 5, Buhl Jr., M. L., A Simple Mode-Shape Generator for Both Towers and Rotating Blades, NWTC Design Codes (Modes), URL: modes/, Last modified August 16, 2; accessed September 26, Laino, D. J. and Hansen, A. C., User's Guide to the Wind Turbine Dynamics Computer Program YawDyn, Salt Lake City, UT: University of Utah, Prepared for the National Renewable Energy Laboratory under Subcontract No. TCX , January Harman, C. R., PROPX: Definitions, Derivations, and Data Flow, Corvalis, OR: Oregon State University, Aug Peters, D. A. and He, C. J., Correlation of Measured Induced Velocities with a Finite-State Wake Model, Journal of American Helicopter Society, July Glauert, H., Airplane Propellers, Aerodynamic Theory, edited by W. F. Durand, Berlin, Julius Springer, Leishman, J. G., and Beddoes, T. S., A Generalized Model for Airfoil Unsteady Behavior and Dynamic Stall Using the Indicial Method, Proceedings from the 42nd Annual Forum of the American Helicopter Society, Washington, DC, 1986, pp Leishman, J. G. and Beddoes, T. S., A Semi-Empirical Model for Dynamic Stall, Journal of the American Helicopter Society, Vol. 34, No. 3, 1989, pp Leishman, J. G., Modeling Sweep Effects on Dynamic Stall, Journal of the American Helicopter Society, Vol. 34, No. 3, 1989, pp Jonkman, B. J. and Buhl, M. L., Jr. TurbSim User's Guide, NREL/EL , Golden, CO: National Renewable Energy Laboratory, September Hand, M. M., Simms, D. A., Fingersh, L. J., Jager, D. W., Cotrell, J. R., Schreck, S., and Larwood, S. M., Unsteady Aerodynamics Experiment Phase VI: Wind Tunnel Test Configurations and Available Data Campaigns, NREL/TP , Golden, CO: National Renewable Energy Laboratory, December Malcolm, D. J. and Hansen, A. C., WindPACT Turbine Rotor Design Study, NREL/SR , Golden, CO: National Renewable Energy Laboratory, Aug International Electrotechnical Commission (TC88), Wind Turbine Generator Systems Part 1: Safety Requirements, Second Edition, IEC 614-1, Geneva: International Electrotechnical Commission, February

17 REPORT DOCUMENTATION PAGE Form Approved OMB No The public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing the burden, to Department of Defense, Executive Services and Communications Directorate (74-188). Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ORGANIZATION. 1. REPORT DATE (DD-MM-YYYY) January REPORT TYPE Conference paper 4. TITLE AND SUBTITLE A Comparison of Wind Turbine Aeroelastic Codes Used for Certification: Preprint 3. DATES COVERED (From - To) January 9-12, 26 5a. CONTRACT NUMBER DE-AC36-99-GO1337 5b. GRANT NUMBER 5c. PROGRAM ELEMENT NUMBER 6. AUTHOR(S) M.L. Buhl, Jr. and A. Manjock 5d. PROJECT NUMBER NREL/CP e. TASK NUMBER WER f. WORK UNIT NUMBER 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) National Renewable Energy Laboratory 1617 Cole Blvd. Golden, CO PERFORMING ORGANIZATION REPORT NUMBER NREL/CP SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 1. SPONSOR/MONITOR'S ACRONYM(S) NREL 11. SPONSORING/MONITORING AGENCY REPORT NUMBER 12. DISTRIBUTION AVAILABILITY STATEMENT National Technical Information Service U.S. Department of Commerce 5285 Port Royal Road Springfield, VA SUPPLEMENTARY NOTES 14. ABSTRACT (Maximum 2 Words) NREL created aeroelastic simulators for horizontal-axis wind turbines accepted by Germanischer Lloyd (GL) WindEnergie GmbH for manufacturers to use for on-shore wind turbine certification. 15. SUBJECT TERMS turbine certification; design codes; ; 16. SECURITY CLASSIFICATION OF: 17. LIMITATION 18. NUMBER 19a. NAME OF RESPONSIBLE PERSON OF ABSTRACT OF PAGES a. REPORT b. ABSTRACT c. THIS PAGE Unclassified Unclassified Unclassified UL 19b. TELEPHONE NUMBER (Include area code) Standard Form 298 (Rev. 8/98) Prescribed by ANSI Std. Z39.18 F1147-E(12/24)